Iron doping in monolayer MoS2 should produce n-type conductivity. Iron substituting for molybdenum introduces electrons. The isolated defect calculation is straightforward, and the prediction is unambiguous.
Experiments show p-type conductivity. The carriers go the wrong way.
Quan and colleagues found the cause: three Fe atoms substituting for Mo arrange themselves into an equilateral triangle around a central sulfur atom. This 3FeMo-S defect associate — a self-organized geometric motif — flips the carrier polarity. The individual dopants behave as predicted. The collective structure does not.
The mechanism is specific. The triangular arrangement creates a local electronic environment that traps the electrons the individual dopants would donate, while simultaneously generating hole states that dominate the transport. The geometry is not incidental. An equilateral triangle has the symmetry to produce the particular electronic structure that reverses the effect. Two Fe atoms near a sulfur don't do it. Four don't do it. Three, in a triangle, does.
The standard assumption in semiconductor doping is that dopant behavior equals isolated-dopant behavior scaled by concentration. This works when dopants are dilute and randomly placed. At higher concentrations, dopants interact. The interaction can enhance, suppress, or — as here — reverse the isolated behavior. The reversal requires a specific geometric arrangement, which means it depends on dopant self-organization kinetics as much as on electronic structure.
This makes doping at finite concentration a fundamentally different problem from doping at infinite dilution. The relevant physics is not the dopant's identity but its social geometry — how it arranges relative to its neighbors. The same atom, in the same host, produces opposite effects depending on whether it finds itself isolated or triangulated.
The prediction from isolated-defect theory was correct. It was also irrelevant to the actual material, where dopants don't stay isolated.